MINFLUX nanometer-scale 3D imaging and microsecond-range tracking on a common fluorescence microscope

The recently introduced minimal photon fluxes (MINFLUX) concept pushed the resolution of fluorescence microscopy to molecular dimensions. Initial demonstrations relied on custom made, specialized microscopes, raising the question of the method’s general availability. Here, we show that MINFLUX implemented with a standard microscope stand can attain 1–3 nm resolution in three dimensions, rendering fluorescence microscopy with molecule-scale resolution widely applicable. Advances, such as synchronized electro-optical and galvanometric beam steering and a stabilization that locks the sample position to sub-nanometer precision with respect to the stand, ensure nanometer-precise and accurate real-time localization of individually activated fluorophores. In our MINFLUX imaging of cell- and neurobiological samples, ~800 detected photons suffice to attain a localization precision of 2.2 nm, whereas ~2500 photons yield precisions <1 nm (standard deviation). We further demonstrate 3D imaging with localization precision of ~2.4 nm in the focal plane and ~1.9 nm along the optic axis. Localizing with a precision of <20 nm within ~100 µs, we establish this spatio-temporal resolution in single fluorophore tracking and apply it to the diffusion of single labeled lipids in lipid-bilayer model membranes.

. Overview of the stabilization unit. The epi-fluorescence beam path and laser safety equipment (fast safety shutter, IR beam block and beam shield) are visible along the excitation and detection beam path of the stabilization unit. A tight cable routing blocks the view of the stabilization camera. Supplementary Fig. S2. Impact of TCP shape on estimator performance. Simulated molecules at positions (crosses) within the FOV and the distributions (red) of their respective estimate represented by their mean (dot) and confidence intervals (ellipsoids) for triangular (a, c) and hexagonal (b, d) TCPs with diameters L of 300 nm (a,b) and 40 nm (c,d). Real-time unbiasing was applied to the radial coordinate only. The benefit of the improved angular uniformity of the localization precision (i.e., low anisotropy in estimator bias) of a hexagonal TCP becomes particularly noteworthy for emitters that reside at the periphery of a TCP with a large diameter. Source data are provided as a Source Data file. Scale bars: 100 nm (a,b), 10 nm (c,d).
Supplementary Fig. S3. Photon dependence of the localization precision. Standard deviations of groups of ≥4 successive localizations obtained from single-fluorophore photon emission bursts establish the localization precision (data from the Nup96 recording shown in Fig. 3d). The localizations are derived from successive and separate groups of Nfin photons from each burst, from the final MINFLUX iterations. a, Histograms of x,y for different photon numbers Nfin. b, Median of x,y vs. mean of photons per aggregate Nagg (log-log representation). c,d Data as in a,b for the spectrin recording of Fig. 3f. e,f Data as in a,b for the PMP70 recording of Fig. 3g. Source data are provided as a Source Data file.

Supplementary Fig. S4. Microscope precision limit measured with a 40 nm bead sample.
Plotted is the mean standard variation of bead localizations along the x and y direction and their respective one-sigma confidence intervals (bounded by solid and dashed lines for x-and y direction, respectively), versus the number of photons used per localization for pattern diameters L of 150 nm and 40 nm. Source data are provided as a Source Data file.

Supplementary Note 1. Microscope precision limit.
The theoretical precision limit given by the Cramér-Rao lower bound (CRLB) scales down to arbitrarily small values in proportion to 1/sqrt(N), with N denoting the photons detected per localization. The practical limit of the precision is finite though, as it is influenced by the positioning noise of the sample stabilization and the (galvanometer) beam steering. Therefore, in order to characterize the performance of the microscope we measured the spread of repeated localizations using a "good-natured" technical sample. Supplementary Fig. S4 summarizes the localization data that we obtained from optically stabilized (Fig. 1) measurements of 40 nm fluorescent beads at a photon detection rate of ~450 kHz. For varying L values, we recorded about 4•10 6 photons from an individual bead, distributed over >23 independent acquisitions. To estimate the localization error, we binned these photon traces into groups of about N sequential photons, estimated the bead position from each group and calculated the standard deviation of the positions that we obtained from each trace.
The plot of the mean sigma versus the number N of photons used per localization scales with the respective CRLB down to a precision of ~1 nm, after which the plots start to flatten out. This result indicates an effective noise floor and a limit for the obtainable precision. For N = 10 5 photons and L = 40 nm this precision limit is close to 0.2 nm. Histogram of track durations of fluorophores coupled to lipids diffusing in a supported lipid bilayer acquired in MINFLUX tracking experiments using the enhanced temporal acquisition scheme (compare Fig. 4 f, g and h). The data shown is filtered for tracks with mean localization rates ≥6.67 kHz (≤150 µs in between consecutive localizations) and ≤250 kHz mean count rate. Source data are provided as a Source Data file.

Supplementary Note 2. Calibration of the tandem scanner.
The tandem scanner of the microscope steers the probe beam to a target position inside the sample by two crossed electro-optical deflectors (EODs) followed by a galvanometric quadscanner. To address a specific position inside the focal plane, the coordinate frames of these two systems need to be registered with respect to each other. We solved this by recording four scanning images of the same fluorescent bead, with an open pinhole and with a different pair of voltages Note that any beam displacement by the EOD unit will shift the bead position in the scanning image (which is indexed by the galvo axis coordinates) to the opposite direction, and hence generates a sign change in the expression for GrEV above). Covering a larger grid area with more sampling points would additionally allow to measure and compensate non-linearities in the beam displacement generated by the EODs. In practice however, the linear approach sufficed, since the maximum voltages (±150 V) applied during our experiments are moderate compared to the ±500 V limit of the EODs used.